Merge changes from topic '8bit-hbd-idct'

* changes:
  *idct*_neon.c: add missing rtcd include
  idct,msa/neon: exclude idct files from hbd build
  *rtcd_defs.pl: remove empty specialize calls

vp8/common/arm/neon/dc_only_idct_add_neon.c

@@ -10,6 +10,8 @@
 
 #include <arm_neon.h>
 
+#include "./vp8_rtcd.h"
+
 void vp8_dc_only_idct_add_neon(int16_t input_dc, unsigned char *pred_ptr,
                                int pred_stride, unsigned char *dst_ptr,
                                int dst_stride) {

vp8/common/arm/neon/dequant_idct_neon.c

@@ -10,6 +10,8 @@
 
 #include <arm_neon.h>
 
+#include "./vp8_rtcd.h"
+
 static const int16_t cospi8sqrt2minus1 = 20091;
 // 35468 exceeds INT16_MAX and gets converted to a negative number. Because of
 // the way it is used in vqdmulh, where the result is doubled, it can be divided

vp8/common/arm/neon/shortidct4x4llm_neon.c

@@ -10,6 +10,8 @@
 
 #include <arm_neon.h>
 
+#include "./vp8_rtcd.h"
+
 static const int16_t cospi8sqrt2minus1 = 20091;
 // 35468 exceeds INT16_MAX and gets converted to a negative number. Because of
 // the way it is used in vqdmulh, where the result is doubled, it can be divided

vp9/common/vp9_rtcd_defs.pl

@@ -49,13 +49,10 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   # Force C versions if CONFIG_EMULATE_HARDWARE is 1
   if (vpx_config("CONFIG_EMULATE_HARDWARE") eq "yes") {
     add_proto qw/void vp9_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
-    specialize qw/vp9_iht4x4_16_add/;
 
     add_proto qw/void vp9_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
-    specialize qw/vp9_iht8x8_64_add/;
 
     add_proto qw/void vp9_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, int tx_type";
-    specialize qw/vp9_iht16x16_256_add/;
   } else {
     add_proto qw/void vp9_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
     specialize qw/vp9_iht4x4_16_add sse2/;
@@ -70,13 +67,10 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   # Force C versions if CONFIG_EMULATE_HARDWARE is 1
   if (vpx_config("CONFIG_EMULATE_HARDWARE") eq "yes") {
     add_proto qw/void vp9_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
-    specialize qw/vp9_iht4x4_16_add/;
 
     add_proto qw/void vp9_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
-    specialize qw/vp9_iht8x8_64_add/;
 
     add_proto qw/void vp9_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, int tx_type";
-    specialize qw/vp9_iht16x16_256_add/;
   } else {
     add_proto qw/void vp9_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
     specialize qw/vp9_iht4x4_16_add sse2 neon dspr2 msa/;
@@ -96,13 +90,10 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   #
   if (vpx_config("CONFIG_VP9_POSTPROC") eq "yes") {
     add_proto qw/void vp9_highbd_mbpost_proc_down/, "uint16_t *dst, int pitch, int rows, int cols, int flimit";
-    specialize qw/vp9_highbd_mbpost_proc_down/;
 
     add_proto qw/void vp9_highbd_mbpost_proc_across_ip/, "uint16_t *src, int pitch, int rows, int cols, int flimit";
-    specialize qw/vp9_highbd_mbpost_proc_across_ip/;
 
     add_proto qw/void vp9_highbd_post_proc_down_and_across/, "const uint16_t *src_ptr, uint16_t *dst_ptr, int src_pixels_per_line, int dst_pixels_per_line, int rows, int cols, int flimit";
-    specialize qw/vp9_highbd_post_proc_down_and_across/;
   }
 
   #
@@ -111,13 +102,10 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   # Note as optimized versions of these functions are added we need to add a check to ensure
   # that when CONFIG_EMULATE_HARDWARE is on, it defaults to the C versions only.
   add_proto qw/void vp9_highbd_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd";
-  specialize qw/vp9_highbd_iht4x4_16_add/;
 
   add_proto qw/void vp9_highbd_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type, int bd";
-  specialize qw/vp9_highbd_iht8x8_64_add/;
 
   add_proto qw/void vp9_highbd_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, int tx_type, int bd";
-  specialize qw/vp9_highbd_iht16x16_256_add/;
 }
 
 #
@@ -137,7 +125,6 @@ if (vpx_config("CONFIG_VP9_TEMPORAL_DENOISING") eq "yes") {
 
 if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   add_proto qw/int64_t vp9_block_error/, "const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz";
-  specialize qw/vp9_block_error/;
 
   add_proto qw/int64_t vp9_highbd_block_error/, "const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz, int bd";
   specialize qw/vp9_highbd_block_error sse2/;
@@ -146,13 +133,10 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vp9_highbd_block_error_8bit sse2 avx/;
 
   add_proto qw/void vp9_quantize_fp/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
-  specialize qw/vp9_quantize_fp/;
 
   add_proto qw/void vp9_quantize_fp_32x32/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
-  specialize qw/vp9_quantize_fp_32x32/;
 
   add_proto qw/void vp9_fdct8x8_quant/, "const int16_t *input, int stride, tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
-  specialize qw/vp9_fdct8x8_quant/;
 } else {
   add_proto qw/int64_t vp9_block_error/, "const tran_low_t *coeff, const tran_low_t *dqcoeff, intptr_t block_size, int64_t *ssz";
   specialize qw/vp9_block_error avx2 msa sse2/;
@@ -217,26 +201,19 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   # ENCODEMB INVOKE
 
   add_proto qw/void vp9_highbd_quantize_fp/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
-  specialize qw/vp9_highbd_quantize_fp/;
 
   add_proto qw/void vp9_highbd_quantize_fp_32x32/, "const tran_low_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, tran_low_t *qcoeff_ptr, tran_low_t *dqcoeff_ptr, const int16_t *dequant_ptr, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan";
-  specialize qw/vp9_highbd_quantize_fp_32x32/;
 
   # fdct functions
   add_proto qw/void vp9_highbd_fht4x4/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
-  specialize qw/vp9_highbd_fht4x4/;
 
   add_proto qw/void vp9_highbd_fht8x8/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
-  specialize qw/vp9_highbd_fht8x8/;
 
   add_proto qw/void vp9_highbd_fht16x16/, "const int16_t *input, tran_low_t *output, int stride, int tx_type";
-  specialize qw/vp9_highbd_fht16x16/;
 
   add_proto qw/void vp9_highbd_fwht4x4/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vp9_highbd_fwht4x4/;
 
   add_proto qw/void vp9_highbd_temporal_filter_apply/, "uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_width, unsigned int block_height, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count";
-  specialize qw/vp9_highbd_temporal_filter_apply/;
 
 }
 # End vp9_high encoder functions

vpx_dsp/arm/idct16x16_1_add_neon.c

@@ -10,10 +10,12 @@
 
 #include <arm_neon.h>
 
+#include "./vpx_dsp_rtcd.h"
 #include "vpx_dsp/inv_txfm.h"
 #include "vpx_ports/mem.h"
 
-void vpx_idct16x16_1_add_neon(int16_t *input, uint8_t *dest, int dest_stride) {
+void vpx_idct16x16_1_add_neon(const tran_low_t *input, uint8_t *dest,
+                              int dest_stride) {
   uint8x8_t d2u8, d3u8, d30u8, d31u8;
   uint64x1_t d2u64, d3u64, d4u64, d5u64;
   uint16x8_t q0u16, q9u16, q10u16, q11u16, q12u16;

vpx_dsp/arm/idct16x16_neon.c

@@ -8,6 +8,7 @@
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
+#include "./vpx_dsp_rtcd.h"
 #include "vpx_dsp/vpx_dsp_common.h"
 
 void vpx_idct16x16_256_add_neon_pass1(const int16_t *input, int16_t *output,

vpx_dsp/arm/idct32x32_1_add_neon.c

@@ -11,7 +11,7 @@
 #include <arm_neon.h>
 
 #include "./vpx_config.h"
-
+#include "./vpx_dsp_rtcd.h"
 #include "vpx_dsp/inv_txfm.h"
 #include "vpx_ports/mem.h"
 
@@ -93,7 +93,8 @@ static INLINE void ST_16x8(uint8_t *d, int d_stride, uint8x16_t *q8u8,
   return;
 }
 
-void vpx_idct32x32_1_add_neon(int16_t *input, uint8_t *dest, int dest_stride) {
+void vpx_idct32x32_1_add_neon(const tran_low_t *input, uint8_t *dest,
+                              int dest_stride) {
   uint8x16_t q0u8, q8u8, q9u8, q10u8, q11u8, q12u8, q13u8, q14u8, q15u8;
   int i, j, dest_stride8;
   uint8_t *d;

vpx_dsp/arm/idct32x32_add_neon.c

@@ -11,6 +11,7 @@
 #include <arm_neon.h>
 
 #include "./vpx_config.h"
+#include "./vpx_dsp_rtcd.h"
 #include "vpx_dsp/arm/transpose_neon.h"
 #include "vpx_dsp/txfm_common.h"
 
@@ -152,8 +153,8 @@ static INLINE void DO_BUTTERFLY(int16x8_t q14s16, int16x8_t q13s16,
   return;
 }
 
-static INLINE void idct32_transpose_pair(int16_t *input, int16_t *t_buf) {
-  int16_t *in;
+static INLINE void idct32_transpose_pair(const int16_t *input, int16_t *t_buf) {
+  const int16_t *in;
   int i;
   const int stride = 32;
   int16x8_t q8s16, q9s16, q10s16, q11s16, q12s16, q13s16, q14s16, q15s16;
@@ -382,7 +383,8 @@ static INLINE void idct32_bands_end_2nd_pass(
   return;
 }
 
-void vpx_idct32x32_1024_add_neon(int16_t *input, uint8_t *dest, int stride) {
+void vpx_idct32x32_1024_add_neon(const tran_low_t *input, uint8_t *dest,
+                                 int stride) {
   int i, idct32_pass_loop;
   int16_t trans_buf[32 * 8];
   int16_t pass1[32 * 32];

vpx_dsp/arm/idct4x4_1_add_neon.c

@@ -10,10 +10,12 @@
 
 #include <arm_neon.h>
 
+#include "./vpx_dsp_rtcd.h"
 #include "vpx_dsp/inv_txfm.h"
 #include "vpx_ports/mem.h"
 
-void vpx_idct4x4_1_add_neon(int16_t *input, uint8_t *dest, int dest_stride) {
+void vpx_idct4x4_1_add_neon(const tran_low_t *input, uint8_t *dest,
+                            int dest_stride) {
   uint8x8_t d6u8;
   uint32x2_t d2u32 = vdup_n_u32(0);
   uint16x8_t q8u16;

vpx_dsp/arm/idct4x4_add_neon.c

@@ -10,7 +10,10 @@
 
 #include <arm_neon.h>
 
-void vpx_idct4x4_16_add_neon(int16_t *input, uint8_t *dest, int dest_stride) {
+#include "./vpx_dsp_rtcd.h"
+
+void vpx_idct4x4_16_add_neon(const tran_low_t *input, uint8_t *dest,
+                             int dest_stride) {
   uint8x8_t d26u8, d27u8;
   uint32x2_t d26u32, d27u32;
   uint16x8_t q8u16, q9u16;

vpx_dsp/arm/idct8x8_1_add_neon.c

@@ -10,10 +10,12 @@
 
 #include <arm_neon.h>
 
+#include "./vpx_dsp_rtcd.h"
 #include "vpx_dsp/inv_txfm.h"
 #include "vpx_ports/mem.h"
 
-void vpx_idct8x8_1_add_neon(int16_t *input, uint8_t *dest, int dest_stride) {
+void vpx_idct8x8_1_add_neon(const tran_low_t *input, uint8_t *dest,
+                            int dest_stride) {
   uint8x8_t d2u8, d3u8, d30u8, d31u8;
   uint64x1_t d2u64, d3u64, d4u64, d5u64;
   uint16x8_t q0u16, q9u16, q10u16, q11u16, q12u16;

vpx_dsp/arm/idct8x8_add_neon.c

@@ -11,6 +11,7 @@
 #include <arm_neon.h>
 
 #include "./vpx_config.h"
+#include "./vpx_dsp_rtcd.h"
 #include "vpx_dsp/arm/transpose_neon.h"
 #include "vpx_dsp/txfm_common.h"
 
@@ -165,7 +166,8 @@ static INLINE void IDCT8x8_1D(int16x8_t *q8s16, int16x8_t *q9s16,
   return;
 }
 
-void vpx_idct8x8_64_add_neon(int16_t *input, uint8_t *dest, int dest_stride) {
+void vpx_idct8x8_64_add_neon(const tran_low_t *input, uint8_t *dest,
+                             int dest_stride) {
   uint8_t *d1, *d2;
   uint8x8_t d0u8, d1u8, d2u8, d3u8;
   uint64x1_t d0u64, d1u64, d2u64, d3u64;
@@ -267,7 +269,8 @@ void vpx_idct8x8_64_add_neon(int16_t *input, uint8_t *dest, int dest_stride) {
   return;
 }
 
-void vpx_idct8x8_12_add_neon(int16_t *input, uint8_t *dest, int dest_stride) {
+void vpx_idct8x8_12_add_neon(const tran_low_t *input, uint8_t *dest,
+                             int dest_stride) {
   uint8_t *d1, *d2;
   uint8x8_t d0u8, d1u8, d2u8, d3u8;
   int16x4_t d10s16, d11s16, d12s16, d13s16, d16s16;

vpx_dsp/vpx_dsp.mk

@@ -196,8 +196,10 @@ ifeq ($(ARCH_X86_64),yes)
 DSP_SRCS-$(HAVE_SSSE3)  += x86/inv_txfm_ssse3_x86_64.asm
 endif  # ARCH_X86_64
 
+DSP_SRCS-$(HAVE_NEON_ASM) += arm/save_reg_neon$(ASM)
+
+ifneq ($(CONFIG_VP9_HIGHBITDEPTH),yes)
 ifeq ($(HAVE_NEON_ASM),yes)
-DSP_SRCS-yes  += arm/save_reg_neon$(ASM)
 DSP_SRCS-yes  += arm/idct4x4_1_add_neon$(ASM)
 DSP_SRCS-yes  += arm/idct4x4_add_neon$(ASM)
 DSP_SRCS-yes  += arm/idct8x8_1_add_neon$(ASM)
@@ -226,7 +228,6 @@ DSP_SRCS-$(HAVE_MSA)   += mips/idct8x8_msa.c
 DSP_SRCS-$(HAVE_MSA)   += mips/idct16x16_msa.c
 DSP_SRCS-$(HAVE_MSA)   += mips/idct32x32_msa.c
 
-ifneq ($(CONFIG_VP9_HIGHBITDEPTH),yes)
 DSP_SRCS-$(HAVE_DSPR2) += mips/inv_txfm_dspr2.h
 DSP_SRCS-$(HAVE_DSPR2) += mips/itrans4_dspr2.c
 DSP_SRCS-$(HAVE_DSPR2) += mips/itrans8_dspr2.c

vpx_dsp/vpx_dsp_rtcd_defs.pl

@@ -29,31 +29,25 @@ add_proto qw/void vpx_d207_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, co
 specialize qw/vpx_d207_predictor_4x4 sse2/;
 
 add_proto qw/void vpx_d207e_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d207e_predictor_4x4/;
 
 add_proto qw/void vpx_d45_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d45_predictor_4x4 neon sse2/;
 
 add_proto qw/void vpx_d45e_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d45e_predictor_4x4/;
 
 add_proto qw/void vpx_d63_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d63_predictor_4x4 ssse3/;
 
 add_proto qw/void vpx_d63e_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d63e_predictor_4x4/;
 
 add_proto qw/void vpx_d63f_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d63f_predictor_4x4/;
 
 add_proto qw/void vpx_h_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_h_predictor_4x4 neon dspr2 msa sse2/;
 
 add_proto qw/void vpx_he_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_he_predictor_4x4/;
 
 add_proto qw/void vpx_d117_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d117_predictor_4x4/;
 
 add_proto qw/void vpx_d135_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d135_predictor_4x4 neon/;
@@ -65,7 +59,6 @@ add_proto qw/void vpx_v_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const
 specialize qw/vpx_v_predictor_4x4 neon msa sse2/;
 
 add_proto qw/void vpx_ve_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_ve_predictor_4x4/;
 
 add_proto qw/void vpx_tm_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_tm_predictor_4x4 neon dspr2 msa sse2/;
@@ -86,28 +79,23 @@ add_proto qw/void vpx_d207_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, co
 specialize qw/vpx_d207_predictor_8x8 ssse3/;
 
 add_proto qw/void vpx_d207e_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d207e_predictor_8x8/;
 
 add_proto qw/void vpx_d45_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d45_predictor_8x8 neon sse2/;
 
 add_proto qw/void vpx_d45e_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d45e_predictor_8x8/;
 
 add_proto qw/void vpx_d63_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d63_predictor_8x8 ssse3/;
 
 add_proto qw/void vpx_d63e_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d63e_predictor_8x8/;
 
 add_proto qw/void vpx_h_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_h_predictor_8x8 neon dspr2 msa sse2/;
 
 add_proto qw/void vpx_d117_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d117_predictor_8x8/;
 
 add_proto qw/void vpx_d135_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d135_predictor_8x8/;
 
 add_proto qw/void vpx_d153_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d153_predictor_8x8 ssse3/;
@@ -134,28 +122,23 @@ add_proto qw/void vpx_d207_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride,
 specialize qw/vpx_d207_predictor_16x16 ssse3/;
 
 add_proto qw/void vpx_d207e_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d207e_predictor_16x16/;
 
 add_proto qw/void vpx_d45_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d45_predictor_16x16 neon ssse3/;
 
 add_proto qw/void vpx_d45e_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d45e_predictor_16x16/;
 
 add_proto qw/void vpx_d63_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d63_predictor_16x16 ssse3/;
 
 add_proto qw/void vpx_d63e_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d63e_predictor_16x16/;
 
 add_proto qw/void vpx_h_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_h_predictor_16x16 neon dspr2 msa sse2/;
 
 add_proto qw/void vpx_d117_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d117_predictor_16x16/;
 
 add_proto qw/void vpx_d135_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d135_predictor_16x16/;
 
 add_proto qw/void vpx_d153_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d153_predictor_16x16 ssse3/;
@@ -182,28 +165,23 @@ add_proto qw/void vpx_d207_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride,
 specialize qw/vpx_d207_predictor_32x32 ssse3/;
 
 add_proto qw/void vpx_d207e_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d207e_predictor_32x32/;
 
 add_proto qw/void vpx_d45_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d45_predictor_32x32 ssse3/;
 
 add_proto qw/void vpx_d45e_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d45e_predictor_32x32/;
 
 add_proto qw/void vpx_d63_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d63_predictor_32x32 ssse3/;
 
 add_proto qw/void vpx_d63e_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d63e_predictor_32x32/;
 
 add_proto qw/void vpx_h_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_h_predictor_32x32 neon msa sse2/;
 
 add_proto qw/void vpx_d117_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d117_predictor_32x32/;
 
 add_proto qw/void vpx_d135_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
-specialize qw/vpx_d135_predictor_32x32/;
 
 add_proto qw/void vpx_d153_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
 specialize qw/vpx_d153_predictor_32x32 ssse3/;
@@ -229,34 +207,24 @@ specialize qw/vpx_dc_128_predictor_32x32 msa neon sse2/;
 # High bitdepth functions
 if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   add_proto qw/void vpx_highbd_d207_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d207_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d207e_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d207e_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d45_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d45_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d45e_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d45e_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d63_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d63_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d63e_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d63e_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_h_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_h_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d117_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d117_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d135_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d135_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d153_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d153_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_v_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
   specialize qw/vpx_highbd_v_predictor_4x4 sse2/;
@@ -268,43 +236,30 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vpx_highbd_dc_predictor_4x4 sse2/;
 
   add_proto qw/void vpx_highbd_dc_top_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_top_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_dc_left_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_left_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_dc_128_predictor_4x4/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_128_predictor_4x4/;
 
   add_proto qw/void vpx_highbd_d207_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d207_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d207e_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d207e_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d45_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d45_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d45e_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d45e_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d63_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d63_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d63e_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d63e_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_h_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_h_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d117_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d117_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d135_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d135_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d153_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d153_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_v_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
   specialize qw/vpx_highbd_v_predictor_8x8 sse2/;
@@ -316,43 +271,30 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vpx_highbd_dc_predictor_8x8 sse2/;;
 
   add_proto qw/void vpx_highbd_dc_top_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_top_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_dc_left_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_left_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_dc_128_predictor_8x8/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_128_predictor_8x8/;
 
   add_proto qw/void vpx_highbd_d207_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d207_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d207e_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d207e_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d45_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d45_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d45e_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d45e_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d63_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d63_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d63e_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d63e_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_h_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_h_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d117_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d117_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d135_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d135_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d153_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d153_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_v_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
   specialize qw/vpx_highbd_v_predictor_16x16 sse2/;
@@ -364,43 +306,30 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vpx_highbd_dc_predictor_16x16 sse2/;
 
   add_proto qw/void vpx_highbd_dc_top_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_top_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_dc_left_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_left_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_dc_128_predictor_16x16/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_128_predictor_16x16/;
 
   add_proto qw/void vpx_highbd_d207_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d207_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_d207e_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d207e_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_d45_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d45_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_d45e_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d45e_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_d63_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d63_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_d63e_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d63e_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_h_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_h_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_d117_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d117_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_d135_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d135_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_d153_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_d153_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_v_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
   specialize qw/vpx_highbd_v_predictor_32x32 sse2/;
@@ -412,13 +341,10 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vpx_highbd_dc_predictor_32x32 sse2/;
 
   add_proto qw/void vpx_highbd_dc_top_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_top_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_dc_left_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_left_predictor_32x32/;
 
   add_proto qw/void vpx_highbd_dc_128_predictor_32x32/, "uint16_t *dst, ptrdiff_t y_stride, const uint16_t *above, const uint16_t *left, int bd";
-  specialize qw/vpx_highbd_dc_128_predictor_32x32/;
 }  # CONFIG_VP9_HIGHBITDEPTH
 
 #
@@ -452,19 +378,14 @@ add_proto qw/void vpx_scaled_2d/, "const uint8_t *src, ptrdiff_t src_stride, uin
 specialize qw/vpx_scaled_2d ssse3/;
 
 add_proto qw/void vpx_scaled_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
-specialize qw/vpx_scaled_horiz/;
 
 add_proto qw/void vpx_scaled_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
-specialize qw/vpx_scaled_vert/;
 
 add_proto qw/void vpx_scaled_avg_2d/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
-specialize qw/vpx_scaled_avg_2d/;
 
 add_proto qw/void vpx_scaled_avg_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
-specialize qw/vpx_scaled_avg_horiz/;
 
 add_proto qw/void vpx_scaled_avg_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h";
-specialize qw/vpx_scaled_avg_vert/;
 
 if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   #
@@ -615,13 +536,11 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vpx_highbd_fdct8x8 sse2/;
 
   add_proto qw/void vpx_highbd_fdct8x8_1/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vpx_highbd_fdct8x8_1/;
 
   add_proto qw/void vpx_highbd_fdct16x16/, "const int16_t *input, tran_low_t *output, int stride";
   specialize qw/vpx_highbd_fdct16x16 sse2/;
 
   add_proto qw/void vpx_highbd_fdct16x16_1/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vpx_highbd_fdct16x16_1/;
 
   add_proto qw/void vpx_highbd_fdct32x32/, "const int16_t *input, tran_low_t *output, int stride";
   specialize qw/vpx_highbd_fdct32x32 sse2/;
@@ -630,7 +549,6 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vpx_highbd_fdct32x32_rd sse2/;
 
   add_proto qw/void vpx_highbd_fdct32x32_1/, "const int16_t *input, tran_low_t *output, int stride";
-  specialize qw/vpx_highbd_fdct32x32_1/;
 } else {
   add_proto qw/void vpx_fdct4x4/, "const int16_t *input, tran_low_t *output, int stride";
   specialize qw/vpx_fdct4x4 sse2 msa/;
@@ -668,87 +586,62 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   # Note as optimized versions of these functions are added we need to add a check to ensure
   # that when CONFIG_EMULATE_HARDWARE is on, it defaults to the C versions only.
   add_proto qw/void vpx_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vpx_iwht4x4_1_add/;
 
   add_proto qw/void vpx_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
   specialize qw/vpx_iwht4x4_16_add sse2/;
 
   add_proto qw/void vpx_highbd_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-  specialize qw/vpx_highbd_idct4x4_1_add/;
 
   add_proto qw/void vpx_highbd_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-  specialize qw/vpx_highbd_idct8x8_1_add/;
 
   add_proto qw/void vpx_highbd_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-  specialize qw/vpx_highbd_idct16x16_1_add/;
 
   add_proto qw/void vpx_highbd_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-  specialize qw/vpx_highbd_idct32x32_1024_add/;
 
   add_proto qw/void vpx_highbd_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-  specialize qw/vpx_highbd_idct32x32_34_add/;
 
   add_proto qw/void vpx_highbd_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
   specialize qw/vpx_highbd_idct32x32_1_add sse2/;
 
   add_proto qw/void vpx_highbd_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-  specialize qw/vpx_highbd_iwht4x4_1_add/;
 
   add_proto qw/void vpx_highbd_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-  specialize qw/vpx_highbd_iwht4x4_16_add/;
 
   # Force C versions if CONFIG_EMULATE_HARDWARE is 1
   if (vpx_config("CONFIG_EMULATE_HARDWARE") eq "yes") {
     add_proto qw/void vpx_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct4x4_16_add/;
 
     add_proto qw/void vpx_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct4x4_1_add/;
 
     add_proto qw/void vpx_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct8x8_64_add/;
 
     add_proto qw/void vpx_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct8x8_12_add/;
 
     add_proto qw/void vpx_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct8x8_1_add/;
 
     add_proto qw/void vpx_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct16x16_256_add/;
 
     add_proto qw/void vpx_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct16x16_10_add/;
 
     add_proto qw/void vpx_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct16x16_1_add/;
 
     add_proto qw/void vpx_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_1024_add/;
 
     add_proto qw/void vpx_idct32x32_135_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_135_add/;
 
     add_proto qw/void vpx_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_34_add/;
 
     add_proto qw/void vpx_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_1_add/;
 
     add_proto qw/void vpx_highbd_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-    specialize qw/vpx_highbd_idct4x4_16_add/;
 
     add_proto qw/void vpx_highbd_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-    specialize qw/vpx_highbd_idct8x8_64_add/;
 
     add_proto qw/void vpx_highbd_idct8x8_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-    specialize qw/vpx_highbd_idct8x8_10_add/;
 
     add_proto qw/void vpx_highbd_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-    specialize qw/vpx_highbd_idct16x16_256_add/;
 
     add_proto qw/void vpx_highbd_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
-    specialize qw/vpx_highbd_idct16x16_10_add/;
   } else {
     add_proto qw/void vpx_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
     specialize qw/vpx_idct4x4_16_add sse2/;
@@ -807,46 +700,32 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   # Force C versions if CONFIG_EMULATE_HARDWARE is 1
   if (vpx_config("CONFIG_EMULATE_HARDWARE") eq "yes") {
     add_proto qw/void vpx_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct4x4_1_add/;
 
     add_proto qw/void vpx_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct4x4_16_add/;
 
     add_proto qw/void vpx_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct8x8_1_add/;
 
     add_proto qw/void vpx_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct8x8_64_add/;
 
     add_proto qw/void vpx_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct8x8_12_add/;
 
     add_proto qw/void vpx_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct16x16_1_add/;
 
     add_proto qw/void vpx_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct16x16_256_add/;
 
     add_proto qw/void vpx_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct16x16_10_add/;
 
     add_proto qw/void vpx_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_1024_add/;
 
     add_proto qw/void vpx_idct32x32_135_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_135_add/;
 
     add_proto qw/void vpx_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_34_add/;
 
     add_proto qw/void vpx_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_idct32x32_1_add/;
 
     add_proto qw/void vpx_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_iwht4x4_1_add/;
 
     add_proto qw/void vpx_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-    specialize qw/vpx_iwht4x4_16_add/;
   } else {
     add_proto qw/void vpx_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
     specialize qw/vpx_idct4x4_1_add sse2 neon dspr2 msa/;
@@ -1153,7 +1032,6 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   # Block subtraction
   #
   add_proto qw/void vpx_highbd_subtract_block/, "int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride, int bd";
-  specialize qw/vpx_highbd_subtract_block/;
 
   #
   # Single block SAD
@@ -1192,20 +1070,15 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vpx_highbd_sad8x4 sse2/;
 
   add_proto qw/unsigned int vpx_highbd_sad4x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-  specialize qw/vpx_highbd_sad4x8/;
 
   add_proto qw/unsigned int vpx_highbd_sad4x4/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride";
-  specialize qw/vpx_highbd_sad4x4/;
 
   #
   # Avg
   #
   add_proto qw/unsigned int vpx_highbd_avg_8x8/, "const uint8_t *, int p";
-  specialize qw/vpx_highbd_avg_8x8/;
   add_proto qw/unsigned int vpx_highbd_avg_4x4/, "const uint8_t *, int p";
-  specialize qw/vpx_highbd_avg_4x4/;
   add_proto qw/void vpx_highbd_minmax_8x8/, "const uint8_t *s, int p, const uint8_t *d, int dp, int *min, int *max";
-  specialize qw/vpx_highbd_minmax_8x8/;
 
   add_proto qw/unsigned int vpx_highbd_sad64x64_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
   specialize qw/vpx_highbd_sad64x64_avg sse2/;
@@ -1241,63 +1114,45 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   specialize qw/vpx_highbd_sad8x4_avg sse2/;
 
   add_proto qw/unsigned int vpx_highbd_sad4x8_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-  specialize qw/vpx_highbd_sad4x8_avg/;
 
   add_proto qw/unsigned int vpx_highbd_sad4x4_avg/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred";
-  specialize qw/vpx_highbd_sad4x4_avg/;
 
   #
   # Multi-block SAD, comparing a reference to N blocks 1 pixel apart horizontally
   #
   # Blocks of 3
   add_proto qw/void vpx_highbd_sad64x64x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad64x64x3/;
 
   add_proto qw/void vpx_highbd_sad32x32x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad32x32x3/;
 
   add_proto qw/void vpx_highbd_sad16x16x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad16x16x3/;
 
   add_proto qw/void vpx_highbd_sad16x8x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad16x8x3/;
 
   add_proto qw/void vpx_highbd_sad8x16x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad8x16x3/;
 
   add_proto qw/void vpx_highbd_sad8x8x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad8x8x3/;
 
   add_proto qw/void vpx_highbd_sad4x4x3/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad4x4x3/;
 
   # Blocks of 8
   add_proto qw/void vpx_highbd_sad64x64x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad64x64x8/;
 
   add_proto qw/void vpx_highbd_sad32x32x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad32x32x8/;
 
   add_proto qw/void vpx_highbd_sad16x16x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad16x16x8/;
 
   add_proto qw/void vpx_highbd_sad16x8x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad16x8x8/;
 
   add_proto qw/void vpx_highbd_sad8x16x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad8x16x8/;
 
   add_proto qw/void vpx_highbd_sad8x8x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad8x8x8/;
 
   add_proto qw/void vpx_highbd_sad8x4x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad8x4x8/;
 
   add_proto qw/void vpx_highbd_sad4x8x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad4x8x8/;
 
   add_proto qw/void vpx_highbd_sad4x4x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array";
-  specialize qw/vpx_highbd_sad4x4x8/;
 
   #
   # Multi-block SAD, comparing a reference to N independent blocks
@@ -1346,7 +1201,6 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   #
   if (vpx_config("CONFIG_INTERNAL_STATS") eq "yes") {
     add_proto qw/void vpx_highbd_ssim_parms_8x8/, "const uint16_t *s, int sp, const uint16_t *r, int rp, uint32_t *sum_s, uint32_t *sum_r, uint32_t *sum_sq_s, uint32_t *sum_sq_r, uint32_t *sum_sxr";
-    specialize qw/vpx_highbd_ssim_parms_8x8/;
   }
 }  # CONFIG_VP9_HIGHBITDEPTH
 }  # CONFIG_ENCODERS